You can use this analysis object to calculate the loudness of a stationary or time-varying sound signal.

The loudness compares the physically measurable strength of the sound (sound pressure) to the loudness perceived by humans as perceived loudness.

Type of input data

The Data tab is where you specify how the input data is to be interpreted. A microphone sensitivity of 50 mV/Pa is used for the conversion between voltage values (V) and sound pressure values (Pa). For more details, refer to Calibration in Acoustics.

Different input data types and event types are available depending on the algorithm selected. A list is available under Loudness.

Supported Types

•The input signal is the third octave spectrum (ISO 532 B, ISO 532-1, ISO 532-2) or octave spectrum (ISO 532 A) of the stationary sound. The third octave spectrum must be a data series or a signal with 28 third octave band levels for frequencies between 25 Hz and 12500 Hz (ISO 532 B, ISO 532-1). The third octave spectrum must be a data series or a signal with 29 third octave band levels for frequencies between 25 Hz and 12500 Hz (ISO 532-2). The octave spectrum must be a data series or a signal with 9 octave band levels for frequencies between 31.5 Hz and 8000 Hz (ISO 532 A).

•The input signal is the measured voltage signal of a stationary sound.

•The input signal is the measured voltage signal of a time-varying sound.

•The input signal is the measured sound pressure signal of a stationary sound.

•The input signal is the measured sound pressure signal of a time-varying sound.

•The input signal is the dichotic third octave spectrum of the stationary sound. This is a data matrix or signal series with two columns which represent the laterally different audible signals. In this case, both signals are considered and produce a loudness value. This input type is only supported by the ISO 532-2 method.

•The input signal is the dichotic voltage signal of a stationary sound. This is a data matrix or signal series with two columns which represent the laterally different audible signals.In this case, both signals are considered and produce a loudness value. This input type is only supported by the ISO 532-2 method.

•The input signal is the dichotic sound pressure signal of a stationary sound. This is a data matrix or signal series with two columns which represent the laterally different audible signals. In this case, both signals are considered and produce a loudness value. This input type is only supported by the ISO 532-2 method.

Calibration

To deduce the loudness from the microphone output voltages recorded, the microphone sensitivity and the gain of the whole signal chain must be taken into consideration. This is done for example by generating a specified sound pressure level at the microphone using a calibrator. This analysis object supports the following working methods:

•If you want to determine the calibration using a calibration measurement, choose Obtain calibration from data set and then specify this calibration measurement as a calibration data set as well as the level of the calibrator (Calibration level). The level must remain constant for at least 4 seconds so that the calibration value can be determined.

If you do not want to have the calibration value calculated every time, press Calibrate. The calibration value is then determined from the data set to be specified and calibration level specified, entered in the Calibration value field, and the calibration mode changes to Fixed (in dB).

•You already know the calibration value in dB required for your microphone. Select Calibration fixed (in dB)  and then specify the Calibration value directly in dB. A calibration value of 0 dB corresponds to a microphone sensitivity of 50 mV/Pa.

•You know the microphone sensitivity in mV/Pa from the calibration certificate. Select Calibration fixed (in mV/PA)  and specify the microphone sensitivity directly in mV/Pa.

For more details, refer to Calibration in Acoustics.

Algorithm

There are four algorithms available for calculating the loudness.

Before the actual loudness calculation is made, an octave and third octave analysis is performed using digital filters in the time domain. The advantage of calculating with time domain filters as opposed to calculating with FFT (Fast Fourier Transform) is the greater amount of accuracy for lower frequencies. For the Zwicker method (ISO 532-1, ISO 532 B), 28 mean third octave levels are calculated in the frequency range of 25 - 12500 Hz.  The Moore-Glasberg method (ISO 532-2) expects a third octave spectrum with 29 third octave band levels for frequencies between 25 Hz and 12500 Hz. For the Stevens method (ISO 532 A), 9 mean octave levels are calculated in the frequency range of 25 - 8000 Hz.  In this case, no frequency weighting takes place.

The algorithms used for the ISO 532-1 and ISO 532-2 methods are oriented on the particular reference implementation, which is described in the standards. The algorithm for the Zwicker method (ISO 532 B) is based on a BASIC program described in the DIN 45631 standard. The ISO 532-1 contains the calculation of the required third octave spectra. For all other methods the required spectra are implicit with the help of the algorithms used in the TimeDomainOctaveAnalysis and SoundLevel FPScript functions.

Two variants of the Zwicker method are available. The ISO 532-1 standard expands on the algorithm (ISO 532 B) already available in earlier versions to include a method for determining the loudness of time-varying sounds, since the old method provided sound values that were too low in this case. To determine the loudness, a third octave spectrum is first determined using a digital filter bank. The Zwicker method then works with several templates that apply to particular third octave level ranges or the level or diffuse sound field forms. The measured third octave levels are entered into these templates and further processed graphically using the principle that the reference areas correspond to specific loudnesses and the total area corresponds to the total loudness.

The Moore-Glasberg method is described in the ISO 532-2 standard. After calculating the third octave spectra, excitation patterns of the frequency groups are calculated on the ERB scale. The specific loudnesses are then calculated from this and are added to the total loudness.

The less used Stevens method (ISO 532 A) attempts to determine the mutual influence of the partial loudness through a formula and using a diagram for assistance.

Sound Field

The result depends on the particular sound field of the sound signal. You can specify whether the sound signal was measured in a free sound field or a diffuse sound field. The Stevens method supports the diffuse sound field only.

Time range to skip at the beginning

To calculate the loudness of stationary sound signals, a time range can be removed at the beginning of the input signal. This is useful, for example, when the input signal contains a calibration measurement at the beginning. This value is ignored in the case of time-varying signals.

Result

The result can be output in various forms:

•Loudness over time. Outputs a signal. The Y component contains the loudness values (in sones). The X component contains the time axis. This result type is useful only for time-varying sound signals, but for reasons of compatibility is also allowed for the old ISO 532-A (Stevens) and ISO 532-B (Zwicker) methods.

•Loudness. Outputs a scalar value. The unit of total loudness used is sone.

•Loudness level.  Outputs a scalar value. The loudness level is determined from the loudness using a conversion formula. The unit used is phon.

•Specific Loudness.  Outputs a signal (stationary sound signals) or a signal series (time-varying sound signals). The specific loudness shows the distribution of loudness over the frequency groups. Th unit of Y component if sone/Bark. The X component in stationary signals contains the tonality in Bark. In time-varying signals the X component contains the time in s and the Z component contains the tonality. The total loudness is the result of specific loudnesses by integration via the tonality.

•Maximum Loudness. Outputs a scalar value. The unit of maximum loudness used is sone. In the case of stationary signals this value is the total loudness.

The Loudness makes a direct statement about how loud a human perceives the sound. A sound that is perceived to be twice as loud obtains twice the loudness value; a sound perceived as half as loud obtains half the loudness value. FlexPro supports two methods for calculating the loudness of stationary sound signals: the Stevens method (ISO 532A) and the Zwicker method (ISO 532B). The Loudness level is a reference standard. It describes which sound pressure level must have a pure tone with a frequency of 1000 Hz so that this is perceived to be just as loud as the sound observed. At this frequency, the sound pressure level (dB SPL) and loudness level (phon) match. A sound with the loudness level of 40 phon corresponds to the loudness of 1 sone. For the Loudness over time the sound levels over time are determined first. The Y component contains the levels in the relevant frequency bands over time. The X component contains the midband frequencies. The Z component contains the time axis. Please note that the settling time is cut off by filtered input signal. The loudness is then determined for each period.

Observed Standards

FPScript Functions Used

AcousticCalibration

Loudness

SoundLevel

TimeDomainOctaveAnalysis

See Also

Sharpness

Acoustics Option

Analysis Objects

Calibration in Acoustics